With its status as a platform company, Nvidia prides itself on enabling all layers of the technology stack – including supporting all phases of the AI lifecycle. For Kanika Atri, a senior director of telecoms at Nvidia, she sees the entire telecom stack being accelerated by AI, particularly as the industry moves towards enabling 6G.

We spoke with Atri about the impact of AI on 6G, the evolution of AI-RAN and how Nvidia will continue to scale its telecoms efforts moving forward.

Delivering the future of AI-led networks

Nvidia continues to make breakthroughs in AI and accelerated computing, supporting most major industries with their digital transformation efforts. This has prompted founder and CEO Jensen Huang to say he predicts the company will make US$1 trillion from AI chips through 2027.

“AI is able to do productive work, and therefore, the inflection point of inference has arrived,” Huang explained at Nvidia GTC this week. “AI now has to think … [it] is going to be much faster than us.”

In this vein, Nvidia’s fourth annual State of AI in Telecommunications survey report found that AI has quickly become the backbone of autonomous networks and AI-native wireless infrastructure – while also unlocking new business and revenue opportunities.

Atri explained how 90% of telcos said they “are already seeing AI impacting their ROI,” which is either “improving revenues or decreasing costs”.

She added: “Every new technology goes through the cycle of invest, adopt and adapt – where they change how they approach new technology. The telcos we can see are currently in the phase between invest and adopt as they begin to see more benefits.”

As AI scales further throughout the entire stack, the telecom industry is starting to become more AI-native and explore more areas of the technology such as network automation.

“Network automation has become the number one use case for AI,” Atri explained. “AI survives on data and the variety, complexity and velocity of data across networks is perfectly suited for AI.”

With networks growing increasingly complex, Atri said AI becomes a great tool to deliver better customer experiences.

“That’s where we’re seeing the adoption of large telco models and adoption of agentic AI,” she said. “Nvidia has been announcing lots of open-source tools and capabilities that telcos can use and build their own autonomous networks with.”

Preparing for faster network transformation

In a blog written by Atri, NVIDIA and Partners Show That Software-Defined AI-RAN Is the Next Wireless Generation, she explored how AI-RAN is the next wireless generation. Notably, that momentum and convergence towards a “common, software-defined foundation” will set the stage for secure, open and AI-native 6G systems.

Nvidia has been ramping up its AI-RAN efforts alongside Nokia, announcing new collaborations with top telecom operators across Europe, Asia and North America. Powered by AI-RAN platforms, the collaborations include T-Mobile US, SoftBank and Indosat Ooredoo Hutchinson.

“With every G cycle (4G, 5G, etc.), there is a new capability introduced that’s usually a combination of both hardware and software,” she said. “But as we step into the AI era, it moves at the pace of software, or software moves at the pace of AI – and the telecom networks need to be prepared for that.”

This all starts from the foundation of software-defined infrastructure, as Atri explained.

“Once you write your radio network as a workload, for example, like a piece of software, it becomes decoupled from the hardware,” she said. “You can then take that same piece of software and run it on any hardware – while, at the same time, being able to add new capabilities without having to upgrade the hardware.”

For Atri, this is a “tectonic” shift in architecture for the industry and what 6G networks will inevitably be built on.

“That is what AI-RAN is about; software-defined, AI-native and open, secure, trusted.”

The road to 6G demands a ‘shared telco vision’

Nvidia has committed to advancing 6G infrastructure, announcing a commitment in February to build the world’s next generation of wireless networks. Alongside a range of telco leaders, including BT Group, Cisco, Deutsche Telekom, Ericsson and Nokia, among others, the shared commitment has been designed to be open, intelligent, resilient and accelerate innovation.

“6G, on the other hand, is born in the AI era,” Atri shared. “We believe 6G will become the fabric for physical AI, not just connecting humans, but connecting intelligence.

“Not only do the platforms we create need to be software-defined to keep pace with innovation, but they also need to be open to allow other industries to innovate on the platform.”

With autonomous networks being adopted more worldwide, national resilience becomes more critical as attack surfaces expand. Atri explained that AI can make the networks more secure and that openness offers more capabilities to handle security risks.

“This innovation creates a platform for 6G to become a new value generation platform for the telecom industry,” she explained.

Looking ahead, Nvidia’s view is that telcos will become AI-native – starting at the infrastructure, operations and services level.

“There are 23 telecom operators worldwide that are now AI service providers,” she added. “They have built AI Factories and are selling AI services beyond GPU-as-a-Service.”

The next evolution is expected to be the distributed infrastructure level, where a new class of applications will demand inferencing that can only be delivered through a distributed network.

In Atri’s mind, telcos are perfectly suited for this next phase, with the expectation that AI will help enable networks that can self-configure, self-optimise, self-assure and self-heal.

“AI-RAN is very much a fundamental architecture that’s going to define this future AI native vision of telcos,” she said. “We believe that telcos have a very important role to play in the AI value chain.

“With this shift to a software-defined infrastructure, it’s going to unlock a completely new set of opportunities.”

Related stories

Nscale & Microsoft partner with Nvidia & Caterpillar to scale new AI data centre

OpenNebula Systems and Nvidia collaborate to deploy AI Factories at scale

Schneider Electric, Nvidia extend data centre partnership to scale AI Factories

Nvidia crushes forecasts with $57bn quarter as AI chips sell out